fmincon end point constraint

[krishna chaitanya koduru]

I have been working on fmincon for dynamic optimisation of temperature profile for batch crystallization problem. I divided the time into equal intervals and temperature(control variable) is assumed to be constant in every interval.Moments are calculated based on temperature. I have an end point constrain on a moment say µ>8.33e9. It is not a constraint on control variable.Please how to input this constraint for fmincon

[Alan Weiss]

As long as mu is a function of the control variables, you can include it
as a nonlinear constraint. In
http://www.mathworks.com/help/toolbox/optim/ug/brhkghv-11.html#brhkghv-16
write

function [c,ceq] = myconstraint(x)
(calculate mu as a function of x)
c = 8.33e9 - mu;
ceq = [];

Alan Weiss
MATLAB mathematical toolbox documentation

[krishna chaitanya koduru]

thanx for the reply but the problem is there is no direct relation fo mu with the control variable T. It's relation with t is complicated . For example mu is a vector of 8 variables and mu(2) depends on mu(1) , mu(3) depends on mu(2) , mu(4) depends on mu(3) and so on till mu(8). All i.e., mu(1),mu(2).....mu(8) depends on T. Hence by varying T , mu is changed. Now , I have a constraint on mu(4) which can't be writted directly w.r.t to T. It is intrincially related. When I wrote that in non-linear constraints C its not working. Now, how the constraint should be given.

P.S.:- Working on the method of moments for calculating optimal temperature profile by assuming the temperature const for each interval and minimising the objective function mu4-mu3. I have end point constraint on mu3 that mu3>8.33e9. How it is applied for fmincon.



Alan Weiss <awe...@mathworks.com> wrote in message <jagptm$t4i$1...@newscl01ah.mathworks.com>...

[Matt J]

"krishna chaitanya koduru" wrote in message <jah1b8$pnk$1...@newscl01ah.mathworks.com>...

>
> thanx for the reply but the problem is there is no direct relation fo mu with the control variable T. It's relation with t is complicated .

=============

There's no reason that should be an issue, as long as the function mu(T) is twice differentiable.

>When I wrote that in non-linear constraints C its not working. Now, how the constraint should be given.

==============

Show us what you wrote.